Nested Unfolding Network for Real-World Concealed Object Segmentation

2025-11-22

AI Summary

Existing deep unfolding network (DUN)-based methods for camouflaged object segmentation (COS) suffer from two key limitations: (1) entangled background estimation and image restoration leading to task interference, and (2) reliance on predefined degradation models, limiting generalizability to real-world scenarios. To address these, we propose the Nested Unfolding Network (NUN), the first DUN-in-DUN architecture that decouples and jointly optimizes restoration and segmentation. NUN integrates vision-language model–guided semantic degradation reasoning with no-reference image quality assessment for prior-free, adaptive restoration. Furthermore, it introduces a reversible foreground-background estimation module and a self-consistency loss to enforce structural coherence. Extensive experiments demonstrate state-of-the-art performance on both clean and degraded image benchmarks, significantly improving segmentation robustness and accuracy under complex real-world conditions.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Deep unfolding networks (DUNs) have recently advanced concealed object segmentation (COS) by modeling segmentation as iterative foreground-background separation. However, existing DUN-based methods (RUN) inherently couple background estimation with image restoration, leading to conflicting objectives and requiring pre-defined degradation types, which are unrealistic in real-world scenarios. To address this, we propose the nested unfolding network (NUN), a unified framework for real-world COS. NUN adopts a DUN-in-DUN design, embedding a degradation-resistant unfolding network (DeRUN) within each stage of a segmentation-oriented unfolding network (SODUN). This design decouples restoration from segmentation while allowing mutual refinement. Guided by a vision-language model (VLM), DeRUN dynamically infers degradation semantics and restores high-quality images without explicit priors, whereas SODUN performs reversible estimation to refine foreground and background. Leveraging the multi-stage nature of unfolding, NUN employs image-quality assessment to select the best DeRUN outputs for subsequent stages, naturally introducing a self-consistency loss that enhances robustness. Extensive experiments show that NUN achieves a leading place on both clean and degraded benchmarks. Code will be released.

Problem

• Decouples image restoration from concealed object segmentation
• Handles unknown real-world degradations without predefined priors
• Resolves conflicting objectives in iterative foreground-background separation

Innovation

• Nested unfolding network decouples restoration from segmentation
• Vision-language model dynamically infers degradation semantics
• Multi-stage quality assessment selects optimal restoration outputs